Here are our guardrails for leveraging AI at tension:

1. We CAN use AI to analyze workshop, meeting, and team discussions and offer suggestions, summaries, and surface themes or findings
2. We CAN use AI to identify technical, operational, or financial issues with ideas, features or functionality we ideate early
3. We CAN use AI to spit out 100 wireframes for a product concept we’re messing around with — but actual client deliverable wires will continue to be hand crafted
4. We CAN use AI to craft testing script, interview guide, and analysis drafts
5. We CAN use AI to create early variations of design concepts — but final designs must be 100% crafted by hand
6. We CAN use AI to write documentation
7. We CAN use AI to create user stories, feature descriptions, etc
8. We CAN use AI for emails, meeting summaries, and response
9. We WILL NOT use AI to create tension or client marketing, messaging, blogs, thought papers or imagery
10. We WILL NOT use AI to deliver anything we do not understand, isn’t based on our own thoughts or ideas, or anything we cannot reverse engineer
11. We WILL NOT use AI without reviewing, fact checking, and revising ourselves
12. We WILL NOT use AI in any way that reduces our quality of work, craftsmanship, or innovation
13. We WILL NOT use AI to generate personas, journeys, or business cases unless it is based on our research, findings and interpretations. Someone else’s end user personas are not our clients end user personas
14. We WILL NOT use AI to make decisions, generate unfounded content, or fill in gaps in client deliverables.

Tension is still innovation for hire, an on-demand think tank. We are working on problems others haven’t solved yet. Why would we let our work be derivative of things that have already been done before?